Skip to main content

​
Introduction

This guide walks you through creating and configuring your first AI agent in the Exulu Intelligence Management Platform. You’ll learn how to set up a basic agent, configure its backend, add API credentials, and control access.

​
Prerequisites

Before creating an agent, ensure you have:
  • A running Exulu IMP instance (see Getting Started)
  • Admin rights (required to create agents)
  • An API key for your chosen LLM provider (e.g., Anthropic, OpenAI, Google Vertex AI)
  • Logged into the dashboard at http://localhost:3000
You must have admin rights to create and manage agents. Contact your system administrator if you don’t have the necessary permissions.

​
Step 1: Navigate to Agents

After logging into your Exulu IMP dashboard:
  1. Look at the left-side navigation panel
  2. Click on Agents in the menu
  3. Click the Create New Agent button
You’ll be taken to the agent creation form.

​
Step 2: Configure Basic Information

Start by providing basic details about your agent:

​
Agent Name

Give your agent a descriptive name. For example:
  • “General Support Agent”
  • “Code Review Assistant”
  • “Customer Service Bot”

​
Description

Add a brief description of what the agent does. This helps other users understand the agent’s purpose.
Choose a clear, descriptive name that reflects the agent’s intended use case. This makes it easier for users to find the right agent.

​
Category

Select or create a category to organize your agents. Common categories include:
  • Support
  • Development
  • Sales
  • Research

​
Step 3: Select an Agent Backend

The backend refers to the underlying ExuluAgent class instance configured in code. This defines which language model and provider your agent uses.

​
Understanding Backends

Agent backends are created by developers and represent specific language model configurations. Each backend includes:
  • Provider - The AI service provider (Anthropic, OpenAI, Google, Azure, etc.)
  • Model - The specific language model (Claude Sonnet 4.5, GPT-4o, etc.)
  • Configuration - Custom settings for the model

​
Default Backends

Exulu IMP includes several pre-configured backends out of the box:
ProviderModelUse Case
AnthropicClaude Sonnet 4.5General purpose, coding, analysis
OpenAIGPT-4oConversational AI, general tasks
OpenAIo1Advanced reasoning tasks
GoogleGemini ProMultimodal tasks
AzureAzure OpenAI modelsEnterprise deployments
Select the backend that best matches your use case. For this example, we’ll use Anthropic Claude Sonnet 4.5.
Custom backends can be created in code by developers. See the ExuluAgent Configuration guide for details.

​
Additional Options

  • Image Generation: Toggle this on if you want the agent to generate images (requires appropriate backend support)
  • Multimodal Support: Some backends support image inputs automatically

​
Step 4: Set System Instructions (Optional)

System instructions provide global guidance to your agent that applies across all user sessions.

​
What to Include

System instructions can define:
  • The agent’s role and persona
  • Response style and tone
  • General guidelines for tool usage
  • Domain-specific knowledge or constraints
  • Output formatting preferences

​
Example System Instructions

You are a helpful customer support agent for Acme Corporation.
Always be polite, professional, and concise in your responses.
When users ask about account issues, use the account lookup tool
before providing information. If you cannot help, escalate to
a human agent.
Keep system instructions general. User-specific customizations can be added at the session level.
You can leave this empty for now and add it later as needed.

​
Step 5: Configure API Key

Most LLM providers require an API key for authentication. Exulu IMP stores these securely using encrypted variables.

​
Creating an API Key Variable

If you haven’t already created a variable for your API key:
  1. Navigate to Variables in the left sidebar
  2. Click Add Variable
  3. Configure the variable:
    • Name: Give it a descriptive name like ANTHROPIC_API_KEY or OPENAI_API_KEY
    • Value: Paste your API key from the provider
    • Encrypt: IMPORTANT - Check this box
  4. Click Save
Always enable encryption for API keys. The system will reject unencrypted API keys for security reasons.

​
Assigning the API Key to Your Agent

Back in the agent configuration:
  1. Find the API Key dropdown
  2. Select the variable you just created
  3. The agent will now use this key when making API calls
You can obtain API keys from your provider’s dashboard:Anthropic: console.anthropic.comOpenAI: platform.openai.comGoogle: console.cloud.google.com

​
Step 6: Activate the Agent

Before users can interact with your agent:
  1. Toggle the Active switch to ON
  2. Click Save to create the agent
Your agent is now live and ready for testing!
Your agent has been created and activated successfully!

​
Step 7: Test Your Agent

Let’s verify the agent works correctly:
  1. Navigate to Chat in the left sidebar
  2. Click Start New Session
  3. Select your newly created agent from the dropdown
  4. Type a test message like “Good morning” or “Hello, how can you help me?”
  5. The agent should respond appropriately
If you receive a response, congratulations - your agent is working!
This is a basic agent without tools or context. Continue to the next guides to learn about adding tools and knowledge bases.

​
Step 8: Configure Access Control

Control who can use your agent with granular access settings.

​
Access Control Options

Navigate to the Access Control tab on the agent editing page. You have several options:

​
1. Private

  • Only you can see and use this agent
  • Best for personal testing or development

​
2. Specific Users

  • Share with selected individual users
  • Good for team collaboration or beta testing

​
3. Specific Roles

  • Share with users who have certain roles (e.g., “Customer Support”, “Engineering”)
  • Ideal for department-specific agents

​
4. Public

  • Available to all users in your organization
  • Best for widely-used general-purpose agents

​
5. Project-Scoped

  • Share within specific projects
  • Useful for project-specific workflows

​
Setting Access Control

  1. Select your preferred access level
  2. If using Specific Users or Specific Roles, choose from the available options
  3. Click Save to apply the settings
Access control can be changed at any time. Start with private access during testing, then expand as needed.

​
Next Steps

Now that you have a basic agent running, you can enhance it with:

​
Add Tools

Give your agent capabilities like web search, code execution, or database queries:

​
Add Knowledge Bases

Provide domain-specific context through vector databases:

​
Create Workflows

Automate recurring tasks with scheduled workflows:

​
Advanced Agent Configuration

Customize behavior, prompts, and advanced settings:

​
Troubleshooting

​
Agent Won’t Respond

If your agent doesn’t respond or throws errors:
  • Check API Key: Verify the variable is encrypted and contains a valid key
  • Check Provider Status: Ensure the LLM provider’s API is operational
  • Check Logs: Look at backend logs for error messages
  • Verify Activation: Ensure the agent is toggled to “Active"

​
"Permission Denied” Errors

If you can’t create agents:
  • Verify you have admin rights in your user profile
  • Contact your system administrator for permission elevation

​
API Key Not Appearing in Dropdown

If your API key variable doesn’t show up:
  • Ensure you saved the variable
  • Refresh the agent configuration page
  • Verify the variable is properly encrypted

​
Agent Selection Not Available in Chat

If you don’t see your agent in chat:
  • Check access control settings
  • Ensure the agent is activated
  • Verify you have permission to use the agent

​
Best Practices

​
Naming Conventions

  • Use descriptive, clear names
  • Include the model type if you have multiple similar agents
  • Example: “Support Agent (Claude 4.5)” vs “Support Agent (GPT-4)“

​
System Instructions

  • Keep instructions concise and focused
  • Test different instruction sets to find what works best
  • Version your instructions and track changes

​
API Key Management

  • Never share API keys
  • Use separate keys for development and production
  • Rotate keys periodically for security
  • Monitor usage to avoid unexpected charges

​
Access Control Strategy

  • Start with private access during development
  • Gradually expand to specific users for testing
  • Only make public after thorough testing
  • Regularly review who has access

​
Testing

  • Test with a variety of prompts
  • Include edge cases and error scenarios
  • Verify behavior matches expectations
  • Get feedback from actual users before wider rollout

​
Summary

You’ve successfully created your first AI agent in Exulu IMP! You learned how to:
  • Navigate to the agent creation interface
  • Configure basic agent information
  • Select an appropriate backend (LLM provider and model)
  • Securely store and assign API keys
  • Activate and test your agent
  • Configure access control settings
Your agent is now ready to use, but this is just the beginning. Explore tools, knowledge bases, and workflows to create truly powerful AI assistants tailored to your specific needs.